grain-size sorting within river bars in relation to ... · pdf fileloughborough university...
TRANSCRIPT
Loughborough UniversityInstitutional Repository
Grain-size sorting withinriver bars in relation to
downstream fining along awandering channel
This item was submitted to Loughborough University's Institutional Repositoryby the/an author.
Citation: RICE, S.P. and CHURCH, M., 2010. Grain-size sorting within riverbars in relation to downstream fining along a wandering channel. Sedimentol-ogy, 57 (1), pp. 232 - 251.
Additional Information:
• This article was submitted for publication in the journal, Sedimentology[ c© The Authors. Journal compilation c© International Association of Sed-imentologists]. The definitive version is published by John Wiley & Sonsand is available at: http://dx.doi.org/10.1111/j.1365-3091.2009.01108.x
Metadata Record: https://dspace.lboro.ac.uk/2134/13029
Version: Submitted for publication
Publisher: John Wiley & Sons, Ltd. c© The Authors. Journal compilation c©International Association of Sedimentologists
Please cite the published version.
This item was submitted to Loughborough’s Institutional Repository (https://dspace.lboro.ac.uk/) by the author and is made available under the
following Creative Commons Licence conditions.
For the full text of this licence, please go to: http://creativecommons.org/licenses/by-nc-nd/2.5/
1
Scales of variability in bed material gravels: Sorting across river
bars in relation to downstream fining
STEPHEN P. RICE1 and MICHAEL CHURCH2
1. Department of Geography, Loughborough University, Loughborough, Leicestershire, LE11
3TU, UK. ([email protected])*.
2. Department of Geography, The University of British Columbia, Vancouver, British
Columbia, Canada V6T 1Z2.
Submitted to Sedimentology, March, 2009
Short running title: Bar-scale sediment sorting
* corresponding author
2
ABSTRACT
The grain-size variability of riverbed gravels at bar scales is poorly understood, as are the
relations between variability at this scale and at reach and river scales. Surface and subsurface
grain size distributions were therefore examined at reach, bar and bedform scales along lower
Fraser River, British Columbia, Canada. Grain-size variations within compound bars are
conditioned by longitudinal position, elevation and morphological setting. Surface and
subsurface sediments tend to decrease in median size from bar head to bar tail by 33% and
17%, respectively. Higher elevations attract smaller particle sizes because of reduced flow
competence during high stages. Unit bars have surface sediments that are finer and better
sorted than the bed materials in bar-top channels and along the main bar edges. Secondary
unit bars tend to have a lower sand content than other features, a consequence of sediment
resorting. Individual unit bars and gravel-sheets exhibit streamwise grain-size fining and lee-
side sand deposition. Through time, morphological adjustments, even significant amounts of
cut and fill, do not ipso facto cause changes in surface grain sizes, yet sediment characteristics
can change without any significant morphological adjustment taking place. At the reach-scale
there is a clear downstream fining trend, but local variability is consistently high due to
within-bar differences that result from sorting by position, elevation and morphology. The
surface median grain size range on individual bars is, on average, 25% of that along the entire
50 km reach but is 68% on one bar. While the overall fining trend yields a downstream
change in surface median size of 0.76 mm km-1, the average value for head-to-tail size
reduction on individual bars is 6.3 mm km-1, an order of magnitude difference that highlights
the effectiveness of bar-scale sorting processes in gravel-bed rivers. We discuss possibilities
for modelling bar-scale variability and the interaction of the different controls that are
identified.
(307 words)
Keywords: compound bar, fluvial sediment, grain size, gravel sheet, unit bar, wandering
channel
3
INTRODUCTION
A large body of work has revealed how bed material grain-size varies in gravel-bed rivers at
reach and river-length scales and has sought to explain the principal spatial features, including
downstream fining (Sternberg, 1875; Church and Kellerhals, 1978) and the gravel-sand
transition (Yatsu, 1955; Sambrook Smith and Ferguson, 1995). In comparison, the variability
of river-bed grain size at bar scales has been the subject of fewer systematic measurement
programs and modelling efforts despite the casual observation that many have made, that bar-
scale sorting typically produces impressive within-bar grain size differences. Church and
Kellerhals’ (1978) analysis of replicate Wolman samples collected from 39 bar-head sites on
Peace River, British Columbia showed that within apparently homogeneous textural units on
single bars, between-sample variance was significantly greater than within-sample variance
and that this added significant scatter to downstream (between bar) fining patterns. Bar-scale
sorting and its relation to the larger-scale longitudinal trend have substantial implications for
sedimentary geology and fluvial geomorphology in terms of explaining the character of
alluvial fills and analysing controls on bed load sediment transport and channel hydraulics.
Distributed quantitative measurements have rarely been used to systematically examine grain
size across gravel bars (but see Ashworth and Ferguson, 1986; Lunt and Bridge, 2004) but
facies mapping of bar surfaces (Bluck, 1976; Forbes, 1983; Wolcott and Church, 1991),
remote sensing of bar surface texture (Carbonneau et al., 2004; Chandler et al., 2004; Verdu
et al., 2005) and measurements of grain size along transverse transects across braid bar
complexes (Mosley and Tindale, 1985; Dawson, 1988; Seal and Paola, 1995) reveal the
patchy nature of textural variation. This patchiness arises from sorting of heterogeneous
sediments by imperfectly understood mechanisms operating at various scales (Whiting, 1996;
Powell, 1998). At grain scales, segregation by size occurs at entrainment because of size-
dependent differences in inertia, protrusion (hence drag) and pivot angle, and during
deposition because of size-dependent interactions between the bedload and the underlying bed
surface texture - for example “like-seeks-like” phenomenon (Moss, 1963; Kuenen, 1966;
Clifford et al., 1993), congestion sorting (Iseya and Ikeda, 1987; Whiting et al., 1988) and
particle overpassing (Allen, 1983; Carling, 1990). At larger, bedform scales bedload is sorted
during transport because of size-dependent differences in the response of particles to gravity
effects – so called topographic sorting – (Paola, 1989; Lisle et al., 1991) and to variations in
near-bed forces induced by local flow patterns around bedforms (Brayshaw et al., 1983), in
bends and around bars (Ashworth, 1996a). The operation of processes like these at bar scales
4
may have implications for larger scale sorting at reach scales, between bars. For example, in a
sequence of alternate bars or bar-chute units, upstream trapping of relatively coarse materials
in bar heads and pools may reduce the availability of those sizes downstream, forcing reach-
scale longitudinal fining (Bluck, 1987). On compound bars the situation is further
complicated because the unit bars and unit bar remnants that are the primary elements of
bedform construction and that define an important scale at which many sorting phenomena
operate, are arranged in complex vertical and areal patterns that reflect complicated
depositional and erosional histories.
In this paper we describe and explain aspects of grain size variation across compound bars
and discuss bar-scale patterns in relation to larger-scale downstream trends. First, we examine
how bed material grain size varies with longitudinal position on the bar, with elevation of the
bed surface and with the local morphological setting. Each of these factors is implicated in
previous work as a potential cause of bar-scale variability. For example, the most widely
reported bar-scale sorting phenomenon is the tendency for bar heads to be coarse relative to
bar tails (Leopold and Wolman, 1957; Church, 1972; Bluck, 1974; Lewin, 1976; Ashworth
and Ferguson, 1986; Lunt and Bridge, 2004;) but, although head to tail sorting has been
measured on simple unit bars (Smith, 1974), there has not been any systematic quantification
of head to tail fining on compound bars. Similarly, several investigators have noted the
apparent influence of bar morphology for structuring bar-scale variations of grain size (Bluck,
1976; 1979; Lunt and Bridge, 2004) and drawn basic contrasts between channel and bar
characteristics, but there has been relatively little explicit consideration of how grain size
varies between morphological features or across the complex topography of compound bars.
Second, we examine how grain size varies across unit bar surfaces because previous evidence
is inconsistent. While Smith (1974) observed downstream fining of surface sediments on
seven simple unit bars studied on the Kicking Horse River, British Columbia, Lunt and
Bridge (2004) observed downstream coarsening of sub-armour layer sediments on unit bars of
the Sagavanirktok River, Alaska. Examples of both downstream fining (Livesey et al., 1998)
and downstream coarsening (Iseya and Ikeda, 1987) are apparent in flume studies where
sheets have been observed. Third, we consider the short-term temporal variability of bed
material grain size as compound bars are modified by unit bar accretion and erosion. Finally,
we compare our findings with the evidence for systematic downstream changes in bed
sediment texture and examine the combined effects of bar-scale sorting relative to
5
downstream fining rates. The field site is in the gravel reach of lower Fraser River, British
Columbia, Canada.
FRASER RIVER GRAVEL REACH
The gravel reach extends for approximately 50 km from Laidlaw, where the river emerges
from confinement between valleyside slopes and terraces, to Sumas Mountain, immediately
upstream from the town of Mission, where there is a rapid gravel-sand transition (Figure 1).
Through the gravel reach the river exhibits a wandering style within an active channel zone
between one and two kilometres wide that is partially constrained by flood defences. Near
Laidlaw, at the Agassiz gauge, the river is 512 m wide at mean annual flood stage, has a mean
depth of 6.6 m, a mean velocity of 2.6 m s-1 and a gradient of 4.8 x 10-4. Corresponding values
at Mission are w = 540 m, d = 12.6 m, v = 1.5 m s-1 and S = 5.0 x 10-5 (McLean et al., 1999).
Most bed load transport, deposition and reworking occur during an annual snowmelt freshet
in late spring and early summer. At Mission (Water Survey Canada, gauge 08MH024), the
mean annual flow is 3410 m3 s-1 and the mean annual flood is 9790 m3 s-1 .
Rice et al. (2009) present a morphological typology for compound bars in wandering gravel-
bed rivers that is primarily developed from observations in the gravel reach of Fraser River.
The fundamental morphological unit is the familiar pool-riffle-bar triplet (Bluck, 1976;
Lewin, 1976; Ferguson and Werrity, 1983): deposition grows a bar where flow diverges
across an oblique riffle, while scour creates a pool that is constrained by the facing bank
where flow converges below the riffle front. Several styles of bank-attached lateral or, less
frequently, medial bars are the dominant macroforms (Jackson, 1975). The scale of the river
is such that these bars may be up to 2.5 km in length and may exceed a kilometre in width.
The compound bars record vertical and lateral accretion of unit bars and their subsequent
modification by cut and fill processes in secondary channels, seasonal anabranches and
smaller bar-top channels that are progressively occupied as stage rises during flood events.
Unit bars, built by the stacking of gravelly bedload sheets, are the basic building blocks that
record individual sediment depositing episodes and dominate the process of morphological
change. The characteristic length scale varies between 101 and 103 m. Primary unit bars are
those which deliver sediment to the bar complex from the principal channel whereas
secondary unit bars are the product of dispersal of sediment from the primary accretion sites
into and through the bar complex via secondary channels, seasonal anabranches and smaller
6
bar-top channels. Figure 2 illustrates these various supraplatform features for one bar on
Fraser River (see Rice et al. (2009) for additional illustrations, particularly Figures 6, 7, and 9
therein). Analysis of bar development on decadal timescales shows that compound bar
formation is controlled by shifts in the position of the principal channel and, therefore, loci of
erosion and unit-bar deposition, but long-term histories reveal the potential for long sequences
(approaching 100 years) of persistent accretion.
METHODS
To characterise bar sediments along the gravel reach, 53 surface (Wolman) and 48 subsurface
samples were collected from nineteen compound bars between Wahleach bar near Laidlaw
and Yaalstrick bar close to Mission during the winter of 2000. To evaluate gross within-bar
variability, two or more samples were collected from sites within the upper, middle and lower
thirds of most bars. Bulk samples that met the 1% criterion of Church et al. (1987) were
collected by pooling small sub-samples from across a larger area, e.g. the bar head. Wolman
samples were positioned at a representative site within the larger unit and consisted of
approximately 400 grains collected from a 20 x 5 m grid with 0.5 m spacing.
To investigate bar-scale variability in greater detail, an additional 87 surface samples and four
subsurface samples (in this case, from single pits) were collected from Queens Bar in April
2000 (before the freshet). Samples were located in a quasi-systematic manner to cover the
entire survey area and ensure representation of the major morphological features. Wolman
samples of 360-400 grains were collected from homogeneous units at 41 sites and a photo-
count method (see below) was used at 46 sites. In September 2000 (after the freshet), 21 of
these sampling positions were reoccupied and Wolman samples were again collected. In
September 2004, 17 positions were reoccupied and sampled again, and 20 new sites were
established. On this occasion, 19 Wolman samples were collected and the AGS (automated
grain-sizing) photographic method (Graham et al., 2005a; 2005b) was employed at 18 sites.
Additional surface samples were collected from several other sites where features of interest
were surveyed, including 51 photo-count samples on Spring Bar and Gill Island during
September 2001.
During the collection of Wolman samples, templates were used to sort grains more than 8 mm
in diameter into half-psi size classes. Smaller grains were classified as 4 to 8 mm or as less
7
than 4 mm, which we define as the ‘sand’ fraction. From the Wolman samples we obtained a
percentage sand cover value and could also derive either coarse-fraction percentiles from a
distribution truncated at 4 mm or percentiles for the entire distribution including sand. Clean
and sandy gravels (those obscured by a thin, discontinuous veneer of sand) dominate the
active portions of the bed. Although the sand cover is of interest, it is sometimes of interest to
establish the size distribution of the coarser fraction without the sand component because the
sand veneer is quickly flushed away during floods and it is the gravel component that
ultimately determines the stability of the bar surface and reflects the peak entrainment stresses
at a site.
The photo-count method is based on a simple calibration between particle size and the
number of particles per unit area (Rice and Church, 1998). At 83 sites where Wolman data
were collected (on ten bars along the reach) a vertical photograph was taken of a 0.5 by 0.5 m
quadrat laid down at random within the Wolman grid. After correction for sand coverage and
obscuration by shadow, calibration relations were established between count greater than 4
mm and Wolman-derived D50 and D95 (truncated to exclude the < 4 mm sand fraction). Sites
with more than approximately 5 % exposed sand degraded the quality of the relations and
were excluded. The resulting calibrations yield prediction limits (α = 0.05) for mean D50 and
D95 estimates of ±5 and ±17 mm respectively, and were used to estimate coarse grain
percentiles at 38 sites where only photographs were collected (estimates were not obtained at
8 of the original 46 sites where sand cover was greater than 5%). The AGS method employed
in 2004 uses robust image processing tools to automatically identify and measure surface
grain sizes. Wolman samples were collected at eleven sites and provide a check on the
method’s application on Fraser River. Comparison of Wolman with AGS D50 and D95
estimates revealed no significant bias and root mean square differences of ±3 mm and ±9 mm,
respectively.
In addition to grain size information, bed structure was described using the “loose”,
“underloose”, “normally loose” classification suggested by Church (1978) and the degree of
embeddedness and strength of imbrication were evaluated qualitatively. Finally, high-stage
flow direction was estimated by measuring the orientation of imbricated clasts. Ten
measurements were made at each sediment sampling site.
8
OBSERVATIONS
Down-Bar position and grain size
During sampling in 2000, at least one surface sample was collected from the upstream and
downstream thirds of nine bars, and at least one subsurface sample from the upstream and
downstream thirds of fourteen bars along the reach (in a population of 20 bars). For these
cases, the ratio of upstream to downstream median grain sizes (untruncated samples) was
calculated as Δ50 = D50U / D50D, where the subscripts U and D refer to upstream and
downstream positions respectively. In eight of the nine surface cases and ten of the fourteen
subsurface cases, Δ50 > 1.0 indicating that bed material size tends to be coarser in the
upstream thirds of compound bars than in the downstream thirds. Mean surface and
subsurface Δ50 values are 1.47 and 1.19 respectively, indicating 33% and 17% decreases in
median size from bar head to bar tail. This bar-scale longitudinal differentiation decreases
with distance downstream (Figure 3) presumably because of an overall improvement in bed
material sorting due to downstream fining.
Surface elevation and grain size
Across Queens Bar, sites at higher elevations are associated with smaller maximum particle
sizes (Figure 4), presumably because water depth at high-stage and, therefore, maximum
shear stress decline as elevation increases. General competence considerations suggest that
maximum grain size should be related to water depth by
s
SYDγτ ⋅
= *max (1)
where Dmax is maximum particle size, Y is water depth, S is the energy slope, γs is submerged
specific weight (=1.65) and τ* is Shields’ dimensionless critical shear stress (≈ 0.03 for dis-
entrainment). The D95 values plotted in Figure 4 are for 87 surface samples collected along
the flank and through the main anabranches of Queens bar in April 2000. During moderate
discharge at that time, average water surface slope across the sampled area was 0.0004
yielding a first approximation of Dmax / Y = 0.008 (m m-1), the rate at which maximum grain
size should increase as water depth increases. When expressed in terms of the decrease in
grain size per unit increase in bed surface elevation, the appropriate slope is –0.008. The bulk
of the data plotted in Figure 4 is neatly contained by an envelope curve with this slope,
9
supporting the argument that variations in particle size with elevation are caused by
differences in imposed shear stresses.
Morphological setting and grain size
The 87 surface samples from Queens bar were classified according to their site morphology as
primary unit bars, secondary unit bars, bar-top channels and principal channel edge. The
surface sediment characteristics of each group are reported in Table 1a. Although individual
sites in different categories share common surface characteristics, sorting processes do
produce some significant between-group differences. Primary and secondary unit bars have
mean D50 and D95 values that are very similar and significantly finer than those for the bed
materials in bar-top channels and along the main bar edge (t-tests, p < 0.001; Figure 5A). Unit
bar gravels are marginally better sorted and the difference in sorting between primary unit
bars and erosional bar edges is significant (Mann-Whitney U test, p = 0.033). These basic
distinctions are explained by the greater mobility and simple depositional history of the unit
bar sediments and their higher elevations (t-tests, p < 0.001; Figure 5B), and therefore lower
entrainment stresses at high stage. Some sites on the unit bars were ‘over-loose’ and the
majority of sites were ‘normally’ loose (Table 1a). In contrast, ‘under-loose’ sediments
dominated the bar-top channels (62%) and no sites were classed as ‘over-loose’. This reflects
the relative mobility and ephemeral flow across unit bars in contrast to the structural
consolidation achieved by longer-lived competent flows in the surrounding channels.
Secondary unit bars were the only setting where surface censoring was observed, typically at
those sites close to the edge of a leading avalanche face. This presumably reflects the
development of a steep energy gradient and, therefore, strong hydraulic winnowing of matrix
fines across the avalanche face as stage falls.
Average percentage sand is significantly lower on secondary unit bars than in the other three
settings (Mann-Whitney U tests, p < 0.001), each of which exhibits a wide range of sand
cover from zero to over 48 % (Table 1a). While 67 % of secondary unit bar sites had ‘clean’
surfaces and none were ‘heavily’ embedded, in bar-top channels only 17 % were ‘clean’ and
48 % were ‘heavily’ embedded. The high proportion of sand in bar-top channels is caused by
proximal disconnection as stage falls. Fines settle out of suspension or stall as they move
along the bed to partially or completely veneer the gravel surface with sand and then silt. At
distal locations, where connections to the principal or secondary channel are maintained for
long periods so that sediment laden waters are refreshed, these veneers may be several
10
centimetres thick. In contrast, the low proportion of sand on secondary unit bar surfaces
reflects the high relative position of their surfaces, which means that they are less likely to be
inundated by deep, slack water from which a significant amount of fine sediment can be
deposited. High proportions of sand at some sites on primary unit bars reflect local hydraulic
conditions. For example where a unit is being overridden by a fresh bar the surface of the
lower, older unit is prone to fine sedimentation in the lee of the advancing avalanche face.
Single subsurface samples collected from each type of morphological setting on Queens bar
show that subsurface grain-size characteristics also vary between morphological settings
within bars. These samples provide good evidence that the degree of variability in subsurface
sediment characteristics is similar to that of the corresponding surface sediments (Figure 6A)
and, therefore, that the sorting processes that produce diverse surface textures affect the
associated bulk deposits too. Again, a primary distinction can be drawn between the
characteristics of unit bar and channel samples. Samples from the two unit bars yielded
similar grain-size distributions that were deficient in coarse gravels and cobbles relative to the
bar-top and bar-edge channel samples (Table 1b; Figure 6B). Armour ratios (= surface
D50/subsurface D50) were calculated for these four sites using untruncated surface and
subsurface GSDs. Values are low for the unit bars (= 1.3 in both cases) which reflects their
mobility and ephemeral flow regime. In contrast, the bar-edge site has an armour ratio
approaching 3.0, which reflects surface winnowing of fine sediment and is consistent with the
greater structural consolidation of channel surfaces noted above. The bar-top channel site has
a low armour ratio (< 1) because the surface at the sampling location was heavily embedded.
This is common (14 of 29 bar-top channel sites were heavily embedded) and reflects the
propensity for fine sedimentation from still water or slackening flows as bar-top channels are
isolated from the principle and secondary channels by falling stage during the receding limb
of the freshet.
A further source of variability is the degree of bimodality in the subsurface sediments. Each
of the bulk grain size distributions is bimodal with a gap in the 0.5 mm to 5.0 mm range.
However, bimodality is notably stronger in the two unit bar sediments and is weakest in the
bar edge sample (Figure 6B). This suggests that bimodality is preferentially generated by
active transport and deposition or preferentially restricted in high-energy erosive
environments where sand fractions are more mobile.
11
Grain-size variations across sheets and unit bars
Grain size variations across individual unit bars and the gravel-sheets which build them tend
to exhibit streamwise grain-size fining and lee-side sand deposition (Figure 7). On Spring bar,
bar-head unit-bar accretion occurs across the full channel width under a relatively simple flow
field that is uncomplicated by islands or channel anabranches. Numerous gravel sheets and
stalled unit bars are evident on the bar platform. At each of seven sheet fronts, four
photographic grain-size samples were obtained, two upstream of the crest and two
downstream on the lower, partly over-ridden unit in front. All samples were collected within
10 m of the sheet crests which were, on average, 0.11 m high. Sand cover was greater
downstream in six of the seven cases and, on average, there was 5.8 times more sand area on
the surface downstream than upstream. This lee-side sand cover is discontinuous but reached
33% in one case (Figure 7A). Beneath the sand, the lee-side gravels tend to be coarser than in
the advancing sheet and in all seven cases, gravel sizes on the crest were finer than those
downstream (Figure 7B and C). On average, the ratio of upstream to downstream D50 was
0.70 (range, 0.49 to 0.93). The relatively coarse bed material in the lee of each advancing unit
belongs to the tail of the underlying sheet so that the differences in gravel size across each
crest suggest that there is a streamwise decrease in framework grain size toward the sheet
crest (Figure 7D). It appears that, on Spring bar, streamwise sorting occurs within each
individual sheet as it climbs up onto the back of the underlying unit. Similar streamwise
fining is apparent on the surface of flat-topped unit bars, for example flank bars on Gill Island
where D50 declines from 46 mm to 28 mm in 130 m. Fining is not ubiquitous and where the
flow field is complex grain size variations across unit bars are less consistent. Figure 7E
shows D95 values across two unit bars that were deposited on Queens during the 2002 freshet.
Along high-flow paths indicated by clast orientation, gravels more often fine downstream
(e.g. X to X’) but streamwise coarsening is sometimes apparent (e.g. Y to Y’).
The tendency for streamwise fining is consistent with some previous flume observations
(Livesey et al., 1998) but inconsistent with the observations of Whiting et al. (1988) on Duck
Creek, where gravel sheets had a relatively coarse leading edge. The material in motion on
Duck Creek is much finer than that on Fraser River (D50 ≈ 5 mm) and laboratory observations
have demonstrated the importance of a high sand content for the leading-edge congestion and
over-passing process that is associated with this type of sheet migration (Iseya and Ikeda,
1987; Dietrich et al., 1989). Fining on the Fraser River unit bars might simply reflect the
dominance of size-selective sorting where the bed load grain-size distribution is generally
12
more coarse with less sand content and where the relief of the unit bars and sheets is greater
such that depth and therefore competence vary significantly along the streamwise axis. That
is, the streamwise fining reflects the increasing downstream height of sheets that are climbing
onto existing bars.
Morphological development and temporal grain size changes
The emplacement of individual unit bars can be responsible for significant local changes of
grain size (Figure 8). For example, the flank unit bar that formed the bar edge at Calamity
before the 2002 freshet had a mean surface D50 of 32 mm (n = 9). The overriding unit that
became attached during the 2002 freshet consisted of coarser gravels with an average D50 of
44 mm (n = 5), a significant difference (t-test, α = 0.01). Despite such spectacular local events
and the general associations between morphological elements and grain size, analyses of bed
material samples from Queens bar suggest that local grain size modification can also occur in
the absence of topographic change and that topographic change does not necessarily produce
changes in grain size.
In September 2000, grain size samples were collected at 21 positions on Queens bar that had
previously been sampled in April. The freshet that occurred during the intervening period
(maximum daily discharge = 8470 m3 s-1, less than the mean annual flood) caused small-scale
reworking of bar morphology (Rice et al., 2009). Topographic surveys indicate that eight of
the sampled positions had experienced either a significant change in elevation or are likely to
have experienced the passage of a large unit bar. The remaining 13 positions underwent no
significant morphological change and were unlikely to have been affected by the passage of
unit bars. Seven of the same positions were reoccupied in September 2004 and all sites had
undergone significant changes during large-scale modification of the bar in the 2002 freshet
(maximum daily discharge = 11 000 m3 s-1, the fifth largest in the 40-year measured record).
Replicate Wolman samples yield a 95% confidence interval on D50 estimates of ±1.8 mm.
Figure 9 plots D50 for the surveys before and after the 2000 freshet (April and September
2000) and before and after the 2002 freshet (September 2000 and September 2004). The solid
lines indicate the region within which differences could be due to sampling imprecision
(±2.55 mm = √(1.82 +1.82)). It is clear that that there is a good deal of temporal consistency
such that sites are characterised by similar D50 grain-sizes before and after each flood. It is
somewhat surprising in Figure 9A, that there is not a clearer distinction between sites
13
experiencing significant cut and fill or the passage of unit bars and those where no
morphological adjustment was apparent.
On the one hand this suggests that morphological adjustments, even significant amounts of
cut and fill, do not ipso facto cause changes in surface grain sizes and, on the other hand, that
sediment characteristics can change without any significant morphological adjustment taking
place. The former may reflect the relatively well-sorted nature of the sediments on this part of
the river, such that local sorting during cut and fill can produce only small changes in median
size (which is in any case a relatively insensitive measure of change). The latter may indicate
that the magnitude of grain size change due to deep incision or burial beneath active gravel
sheets is no greater than that which can be accomplished in the surface layer by, for example,
winnowing or the discontinuous diffusion of mobile materials.
The limited changes that took place at particular locations on Queens bar are consistent with
the similarity of average bar-scale grain-size characteristics during the study period. When the
mean and variance of the 79, 21 and 17 D50 values obtained on Queens in April 2000,
September 2000 and September 2004 were compared, there were no significant differences
(F- and t-tests, α > 0.10).
The downstream trend
The dominant large-scale trend in many alluvial rivers is downstream fining caused by sorting
and abrasion processes (Russell, 1939; Parker, 1991; Ferguson et al., 1996; Rice, 1999; Lewin
and Brewer, 2002), punctuated by positive grain-size steps at coarse-sediment recruitment
points such as some tributaries (Sternberg, 1875; Miller, 1958; Knighton, 1980; Rice and
Church, 1998). This model may be disrupted to varying degrees by low sediment supply
(Singer, 2008), diffuse lateral sediment sources (Heller et al., 2001; Davey and Lapointe,
2007) and local factors that include hillslope-channel coupling (Rice and Church, 1996),
anthropogenic channel modifications (Surian, 2002) and variations in gravel mobility due to
lithological control of channel morphology (Constantine et al., 2003).
Overall downstream fining is present along the gravel reach of Fraser River, beyond which
there is a transition to sand. In 2000, multiple samples were collected from homogenous
patches on twenty individual bars. Surface D50 declined from approximately 50 to 20 mm
(Figure 10) along the gravel reach, and D90 declined from approximately 100 to 30 mm.
Corresponding values for subsurface sediments were 25 to 10 mm and 90 to 30 mm. Despite
14
these general trends, local variability is consistently high (on the order of 1 to 1.5 psi units or
log2 cycles) and, in common with many published downstream fining datasets, scatter is a
dominant characteristic. It is clear from Figure 10A, where data for individual bars are
differentiated, that much of this scatter is due to within-bar variability rather than between-bar
differences. Indeed, both surface and subsurface sediments can exhibit almost as much
variation within the limits of an individual compound bar as does the entire gravel reach. For
the nine bars where head and tail Wolman samples are available, and where we might
therefore presume to have captured a reasonable amount of the bar-scale textural variability,
the average surface D50 range is 20 mm, the maximum is 51 mm (Wahleach Bar) and the
minimum is 4 mm (Carey Bar). These contrast with the surface D50 range of 90-15 = 75 mm
for the entire set of gravel bars along the 50 km reach. This indicates that the grain size range
on individual bars is, on average, approximately 25% of the overall range in the reach but can
be as high as 68%.
An alternative expression of this pattern takes into account the distance over which sorting is
apparent at reach and bar scales. For the entire reach, the linear least-squares downstream
fining model yields a diminution rate of 0.76 mm km-1. For the nine bars with head and tail
samples (excluding the one bar where the bar tail was coarser) the rate of decline in median
grain size from head to tail ranges between 0.56 mm km-1 at Yaalstrick and 15.08 mm km-1 at
Gill, with an average of 6.31 mm km-1.
To examine temporal stability of the downstream fining trend, grain-size data from 2000 can
be compared with an earlier data set of bar-head samples collected in 1983/84 (n = 53).
Surface (Wolman) samples collected in the earlier program were only 100 clasts in size, but
that is sufficient to return a reliable estimate of the median grain size (Rice and Church,
1996). The comparison of surface materials indicates that bed materials have become finer in
the sub-reach common to both sampling programmes (between river km 108 and 140). Over
that period, mean surface D50 in the sub-reach appears to have declined significantly from 35
to 29 mm (t-test, p= 0.05). Overall fining trends remain consistent, however, with no
significant difference in the regression slopes for linear fining models for the two years
(Ancova, p = 0.83; Figure 10B). There is a possibility that differences between the two
sampling programs in sample protocol with respect to the finest grains may have induced the
result. However, reach-scale changes are presumably dominated by patterns of sediment
15
supply and transport capacity; a long run of below normal freshets between the two sampling
dates may very well have effected such a change.
Combined effects
The question arises whether the sedimentological effects described above act in combination
to affect grain size. Because downstream fining effects impose a systematic difference in
expected mean grain size from bar to bar, we ask whether the bar-scale variations we have
studied may affect that trend locally. Down-bar fining – also a systematic spatial effect –
needs to be averaged away in a representative way in order to study downstream fining from
bar to bar, while grain size variations associated with morphological setting, which vary
spatially within bars, also needs to be averaged. We also know that surface elevation affects
sediment size within bars and we do know bar platform elevation (see Church and Rice, in
review), that is, the elevation of the gravel surface above the main channel bed. Hence, using
the spatially averaged mean grain size for each bar, we may ask whether bar platform
elevation explains any of the substantial scatter of data about the mean downstream trend. We
examine this question by computing the downstream trend for data averaged across individual
bars and then examining the correlation between bar platform elevation and the residuals from
that trend. Surface and subsurface data of 12 bars are available for analysis.
Some results are displayed in Figure 11. To compute the trend line, data of Foster Bar were
excluded. Foster Bar presents a large anomaly, a circumstance that probably is related to a
large gravel extraction that occurred on the bar 4 years before sampling. The residuals do
correlate with bar thickness, except that Big Bar again presents a significant anomaly. Big Bar
is a relatively recently established bar (45 years) that has formed on top of a long diagonal
riffle in the river. The bar tail sits astride the riffle and there is a large, perennially flowing
chute channel below the riffle. With this morphology, there is little bar tail deposition of finer
sediments, which serves to elevate the mean grain size for the bar. For the remaining bars the
correlation of the residuals with bar thickness is significant (p = 0.06; for exploratory
purposes we adopt α = 0.10 as the threshold for a significant result). We tentatively conclude
that bar thickness influences grain size downstream and helps to explain residual scatter about
the reach-scale fining trend. The results displayed in Figure 11 are for surface D50. Variance
reduction in the same analysis is marginally superior for surface D90 and the pattern is
identical, but no significant effect was found for subsurface sediments.
16
Given this effect and the general expectation that bars thicken as they grow, the further
question arises whether bar age is correlated with grain size (in the absence of the fining trend
and other local effects). We know the ages of the bars, but we found that bar age did not
correlate with the residual variance from the downstream fining trend line and conclude that
there is no age effect. This is explained by the observation that once established, bars rapidly
assume a surface elevation that is close to the final elevation (Church and Rice, in review)
such that bar thickness is essentially independent of bar age.
DISCUSSION
We have demonstrated grain size variations in a gravel-bed river at scales varying from
individual sedimentary units (unit bars) within bar, through the dominant bar scale, to whole
reach scale. Within sedimentary units, size gradation is associated with distance along a
flowline and with elevation, themselves correlated so that fining occurs upward and
downstream. Overall, fining is associated with flow divergence and sediment depositional
gradient. Secondary sorting in bar top channels and resedimented units complicates the local
pattern of grain size variation so that a somewhat palimpsestic pattern of repeated local fining
gradients occurs on a compound bar.
An overall pattern of down-bar fining at the bar scale is the strongest individual gradient at
any scale. This gradient is the consequence of sediment sorting associated with steering of the
river current by the developing barform. The major sediment body dominantly grows laterally
(Rice et al., 2009; Figure 8, herein), so the lee-side regions become shadow zones into which
sediment moves over the bar-top, primarily through bar-top channels, or diffuses laterally
from the displaced main current. Predominantly finer sediment is deposited in this shadow
zone. While this bar-scale gradient is strongest, clearly demonstrating that sorting on
individual gravel bars is highly effective, this remains the aspect of bar-scale sedimentation
that is perhaps least well understood in its details.
This explanation has an affinity with Ashworth’s model of bar formation in braided rivers in
terms of the focus on mobile units and their different local transport pathways and sinks
(Ashworth et al. 1992a,b; Ashworth, 1996). In that model, the relatively small impact of
secondary flows on coarse bedload and flow divergence at emergent bar heads is thought to
lead to the concentration of relatively coarse material there, while finer sediments moved
17
laterally by secondary currents are thought to be more easily deflected around bar heads into
distributary channels, where topographic sorting promotes bar-tail deposition. Bluck (1976;
1979) emphasised the stage dependence of depositional positions across a bar platform,
suggesting that coarser bar heads are formed at high stages and finer tails in the lee of these
deposits during falling stages. Thick sandy facies are commonly deposited on Fraser River bar
tails during the receding limb of the annual hydrograph so this process is probably part of the
explanation too. It is unlikely that bar-scale fining reflects the operation of a “turbulence
template” whereby relatively coarse bar head materials generate hydrodynamics that only
relatively coarse bedload grains can tolerate, such that incoming finer grains are rejected and
transported to downstream positions (Bluck,1987; Clifford et al. 1993), because such a
process does not operate at the kilometre scale of the compound bars.
Local sorting across bars and individual bedforms produces grain size variations at the scale
of the channel width that are equivalent to those observed longitudinally over much greater
distances. In practice, this means that downstream fining patterns are revealed most clearly
when sediment variability at a local scale is minimised, usually by limiting sampling to a
particular bar-scale unit (typically the bar head) or by lumping together observations to
provide a spatial average. The usefulness of downstream fining models for prediction is then
limited to reach scale applications, while predictions of bar-scale variability require an
additional term to express the local grain-size range that can be anticipated.
Our analysis of Fraser River sediments suggests two potential means of providing such an
expression. First there is, on average, a 33% reduction in median surface grain size between
bar-head and bar-tail (defined as the upper and lower third of the compound bar surface,
respectively). Bar-scale variability on this river might be approximated by reducing a
prediction of bar-head grain size (D50U) from a downstream fining model by a proportion P (=
0.33 on Fraser River) to estimate the bar-tail grain size and thence the bar-scale range = D50U
(1– P). This simple model and the value of P requires further investigation because other
rivers are likely to exhibit different ratios; for example, the maximum particle size delivered
to bar heads presumably limits the possible magnitude of downstream bar change because the
lower limit is essentially fixed. Indeed, this example suggests a second means of estimating
bar-scale variability that relies on the possibility that the value of P may vary systematically
along a fining trend. We note that along the reach-scale fining sequence, there is a weak
decline in within-bar (between-patch) variability (R2 = 0.55 for the relation between distance
18
downstream and within-bar D50 range; Figure 10A), suggesting that patch textures become
less diverse as a systematic function of position along a fining trend. This is consistent with
the progressive downstream exclusion of coarse grains during downstream fining, leaving
available a reduced mixture of sizes for bar-scale sorting (cf. Bluck, 1987). We hesitate to
define the nature of this relation here, but a larger data set designed to establish within-bar,
between-patch variability along a number of fining sequences, may yield useful empirical
models.
At bar scales, however, there is no equivalent means of modelling textural change through
time, even though such information is important for understanding the details of alluvial
stratigraphy and the impact of river change on lotic habitats. Developing a better spatial and
temporal understanding of textural variability at even finer sub-width scales is also important,
for example, for validating and calibrating 2-D models of channel hydraulics, sediment
transport and patchy lotic habitats. Repeated plane table mapping and sediment-size analysis
of bars on the River Tulla, Scotland by Bluck (1987) revealed that most bars coarsen as they
age, which Bluck suggested was due to in situ modification by the turbulence template
mechanism (Clifford et al., 1993). We do not have a sufficiently long data set to test these
ideas, but our examination of grain size responses to mobile bed events across Queens Bar,
does show that morphological change may or may not cause local grain size changes and that
morphological stability may or may not be associated with grain size change. Moreover, we
found no correlation between bar age and residuals from the downstream fining trend.
CONCLUSIONS
We have demonstrated that sediment grain sizes found on bar surfaces in a 50-km gravel bed
reach of a large wandering gravel-bed river vary systematically with down-bar position, with
morphological setting, and with bar surface elevation. In addition, sediment texture can be
modified by secondary resedimentation on a bar but, on the bars found in this large river,
representative grain size does not necessarily change systematically, at a particular position,
over a period of years that includes repeated seasonal inundation.
The qualitatively well-known down-bar variation is by far the greatest source of variation,
quantitatively larger locally than the overall downstream fining of fluvial gravels. On lower
Fraser River, the difference amounts to an order of magnitude (6.3 mm km-1 vs. 0.76 mm km-
19
1). Its occurrence is mediated by the mutual effect of bar growth and river current steering,
which determines the developing pattern of sedimentation. Local variation in grain size due to
the surface elevation of the bar surface above the main channel bed contributes to the
observed variance in overall downstream fining.
This finding carries important implications for representative sampling of fluvial bed material
grain size along gravel-bed rivers and for both the calibration of 2-D morphodynamic models
and the interpretation of bed material grain size predicted by those models. High bar-scale
variability means that large-scale trends can only be revealed when local-scale variability is
removed; for example, by consistently sampling from a single depositional environment, like
bar-heads. While this is well-established practise amongst most geomorphologists and
sedimentologists dealing with modern sediments, it is clearly difficult to achieve when
examining limited exposures in alluvial fills or lithified gravels. High local variability also
means that single bed material samples cannot be representative of the grain sizes apparent on
a bar or across a channel width. This is problematic when trying to establish grain roughness
parameters in morphodynamic models of gravel-bed rivers and when comparing model
predictions of bed material grain size with field data (e.g., Ferguson and Church, in review).
Notwithstanding the explanations of bar-scale variability described herein, it is clear that the
complexity of sediment sorting across compound bars will continue to hinder the
identification of simple predictive models of bar-scale grain-size. In this context, empirical
information remains important, both for improving understanding of bar-scale sedimentation
and for providing calibration and test data for morphodyamic models. Recent technical
advances in distributed measurement tools (Carbonneau et al., 2005; Graham et al., 2005b;
Verdu et al., 2005) will improve our ability to document the spatial and temporal variability
of river bed sediments at bar and reach scales. Both automated airborne and ground-based
image-analysis systems now make it feasible to collect high-resolution grain size information
over large areas and, where long-term archives of air photography are available, to unlock
histories of grain-size change
Acknowledgments
Work on lower Fraser River is supported by the Natural Sciences and Engineering Research
Council of Canada through a Discovery Grant to M.Church. A host of assistants and
20
colleagues made major contributions in the field and laboratory, including Jim Chandler,
Erica Ellis, Nick Manklow, Laura Rempel, Hamish Weatherley and André Zimmermann. The
1984 sediment data were collected by David McLean.
References
Allen, J.R.L. (1983) Gravel overpassing on humpback bars supplied with mixed sediment - Examples from the Lower Old Red Sandstone, Southern Britain. Sedimentology, 30: 285-294.
Ashworth, P.J. (1996a) Mid-channel bar growth and its relationship to local flow strength and direction. Earth Surface Processes and Landforms, 21: 103-123.
Ashworth, P.J. and Ferguson, R.I. (1986) Interrelationships of channel processes, changes and sediments in a proglacial braided river. Geografiska Annaler, 68 A: 361-371.
Ashworth, P.J., Ferguson, R.I. and Powell, M.D. (1992a) Bedload transport and sorting in braided channels. In: Dynamics of Gravel-bed Rivers (Eds P. Billi, R.D. Hey, C.R. Thorne and P. Tacconi), pp. 497-515. John Wiley & Sons Ltd, Chichester.
Ashworth, P.J., Powell, D.M., Prestegaard, K.L., Ferguson, R.I., Ashmore, P.E. and Paola, C. (1992b) Measurements in a braided river chute and lobe. 2. Sorting of bed load during entrainment, transport, and deposition. Water Resources Research, 28: 1887-1896.
Bluck, B.J. (1974) Structure and depositional properties of some valley sandur deposits in southern Iceland. Sedimentology, 21: 533-554.
Bluck, B.J. (1979) Structure of coarse grained braided stream alluvium. Transactions of the Royal Society of Edinburgh, 70: 181-221.
Bluck, B.J. (1976) Sedimentation in some Scottish rivers of low sinuosity. Transactions of the Royal Society of Edinburgh, 69: 425-455.
Bluck, B.J. (1987) Bed forms and clast size changes in gravel-bed rivers. In: River Channels: Environment and Process (Eds K. Richards), pp. 391. Blackwell, Oxford.
Brayshaw, A.C., Frostick, L.E. and Reid, I. (1983) The hydrodynamics of particle clusters and sediment entrainment in coarse alluvial channels. Sedimentology, 30: 137-143.
Carling, P.A. (1990) Particle over-passing on depth-limited gravel bars. Sedimentology, 37: 345-355.
Chandler, J.H., Rice, S.P. and Church, M. (2004) Colour Aerial Photography for riverbed classification. Archives of the International Society for Photogrammetry and Remote Sensing, XXXV
Church, M. (1972) Baffin Island sandurs: a study of Arctic fluvial processes. Geological Survey of Canada, Bulletin, 216: 208.
21
Church, M. (1978) Paleohydrological reconstructions from a Holocene valleyfill. In: Fluvial Sedimentology (Proceedings, 1st International Symposium on Fluvial Sedimentology) (Eds A.D. Miall), Canadian Society of Petroleum Geologists Memoir 5: 743-772.
Church, M. and Kellerhals, R. 1978. On the statistics of grain size variation along a gravel river. Canadian Journal of Earth Sciences, 15: 1151-1160
Church, M., McLean, D.G. and Wolcott, J.F. (1987) River bed gravels: sampling and analysis. In: Sediment Transport in Gravel Bed Rivers (Eds C.R. Thorne, J.C. Bathurst and R.D. Hey), John Wiley & Sons Ltd, Chichester.
Clifford, N.J., Hardisty, J., French, J.R. and Hart, S. (1993) Downstream variation in bed material characteristics: a turbulence controlled form-process feedback mechansism. In: Braided Rivers; Geological Society Special Publication no. 75 (Eds J.L. Best and C.S. Bristow), pp. 89-104. The Geological Society, London.
Constantine, C.R., Mount, M.F. and Florsheim, J.L. (2003) The effects of longitudinal differences in gravel mobility on the downstream fining pattern in the Cosumnes River, California. Journal of Geology, 111: 233-241.
Davey, C. and Lapointe, M. (2007) Sedimentary links and the spatial organization of Atlantic salmon (Salmo salar) spawning habitat in a Canadian Shield river. Geomorphology, 83: 82-96.
Dawson, M. (1988) Sediment size variation in a braided reach of the Sunwapta River, Alberta, Canada. Earth Surface Processes and Landforms, 13: 599-618.
Dietrich, W.E., Kirchner, J.W., Ikeda, H. and Iseya, F. (1989) Sediment supply and the development of the coarse surface layer in gravel-bedded rivers. Nature, 340: 215-217.
Ferguson, R.I. and Werrity, A. (1983) Bar development and channel changes in the gravelly River Feshie, Scotland. In: Modern and Ancient Fluvial Systems (Eds J.D. Collinson and J. Lewin), pp. 181-193. Blackwell Science, Oxford, UK.
Ferguson, R.I. (2003) Emergence of abrupt gravel to sand transitions along rivers through sorting processes. Geology, 31: 159-162.
Ferguson, R. I. and Church, M. in review. A critical perspective on 1D modelling of river processes: gravel load and aggradation in lower Fraser River. Water Resources Research.
Ferguson, R.I., Hoey, T., Wathen, S. and Werrity, A. (1996) Field evidence for rapid downstream fining of river gravels through selective transport. Geology, 24: 179-182.
Forbes, D.L. (1983) Morphology and sedimentology of a sinuous gravel-bed channel system: lower Babbage River, Yukon coastal plain, Canada. In: Modern and Ancient Fluvial Systems (Eds J.D. Collinson and J. Lewin), pp. 195-206. Blackwell Scientific, Oxford, UK.
Graham, D., Reid, I. and Rice, S. (2005a) Automated Sizing of Coarse-Grained Sediments: Image-Processing Procedures. Mathematical Geology 37, no.1,
Graham, D.J., Rice, S.P. and Reid, I. (2005b) A transferable method for the automated grain sizing of river gravels. Water Resources Research, 41: 1-12.
22
Heller, P.L., Beland, P.E., Humphrey, N.F., Konrad, S.K., Lynds, R.M., McMillan, M.E., Valentine, K.E., Widman, Y.A. and Furbish, D.J. (2001) Paradox of downstream fining and weathering-rind formation in the lower Hoh River, Olympic Peninsula, Washington. Geology, 29: 971-974.
Iseya, F. and Ikeda, H. (1987) Pulsations in bedload transport rates induced by a longitudinal sediment sorting: a flume study using sand and gravel mixtures. Geografiska Annaler, 69 A: 15-27.
Jackson, R.G. and II (1975) Hierarchical attributes and a unifying model of bedforms composed of cohesionless material and produced by shearing flow. Geological Society of America Bulletin, 86: 1523-1533.
Knighton, A.D. (1980) Longitudinal changes in size and sorting of stream bed material in four English Rivers. Geological Society of America Bulletin, 91: 55-62.
Kuenen, P.H. (1966) Experimental turbidite lamination in a circular flume. Journal of Geology, 74: 523-&.
Leopold, L.B. and Wolman, M.G. (1957) River channel patterns - braided, meandering, and straight. US Geological Survey Professional Paper, 282-B: 39-85.
Lewin, J. (1976) Initiation of bed forms and meanders in coarse-grained sediment. Geological Society of America Bulletin, 87: 281-285.
Lewin, J. and Brewer, P.A. (2002) Laboratory simulation of clast abrasion. Earth Surface Processes and Landforms, 27: 145-164.
Lisle, T.E., Ikeda, H. and Iseya, F. (1991) Formation of stationary alternate bars in a steep channel with mixed size sediment: a flume experiment. Earth Surface Processes and Landforms, 16: 463-469.
Livesey, J.R., Bennett, S., Ashworth, P.J. and Best, B.J. (1998) Flow structure, sediment transport and bedform dynamics for a bimodal sediment mixture. In: Gravel-Bed Rivers in the Environment (Eds P.C. Klingeman, R.L. Beschta, P.D. Komar and J.B. Bradley), pp. 149-176. Water Resources Publications, Colorado, USA.
Lunt, I.A. and Bridge, J.S. (2004) Evolution and deposits of a gravelly braid bar, Sagavanirktok River, Alaska. Sedimentology, 51: 415-432.
McLean, D.G., Church, M. and Tassone, B. (1999) Sediment transport along lower Fraser River: 1. Measurements and hydraulic computations. Water Resources Research, 35: 2533-2548.
Miller, J.P. (1958) High mountain streams: effects of geology on channel characteristics and bed material.
Moss, A.J. (1963) Physical Nature of Common Sandy and Pebbly Deposits .2. American Journal of Science, 261: 297-343.
Mosley M.P. and Tindale, R.S. (1985) Sediment variability and bed material sampling in gravel-bed rivers. Earth Surface Processes and Landforms 10: 465–482.
23
Nicholas, A.P., Ashworth, P.J., Kirkby, M.J., Macklin, M.G. and Murray, T. (1995) Sediment slugs: large-scale fluctuations in fluvial sediment transport rates and storage volumes. Progress in Physical Geography, 19: 500-519.
Paola, C. (1989) Topographic sorting. EOS, Transactions of the American Geophysical Union, 70: 332.
Parker, G. (1991) Selective sorting and abrasion of river gravel. I; Theory. Journal of Hydraulic Engineering, 117: 131-149.
Powell, D.M. (1998) Patterns and processes of sediment sorting in gravel-bed rivers. Progress in Physical Geography, 22: 1-32.
Rice, S.P. (1999) The nature and controls on downstream fining within sedimentary links. Journal of Sedimentary Research, 69: 32-39.
Rice, S.P. and Church, M. (1996) Bed material texture in low order streams on the Queen Charlotte Islands, British Columbia. Earth Surface Processes and Landforms, 21: 1-18.
Rice, S.P. and Church, M. (1996). Sampling surficial fluvial gravels: the precision of size distribution percentile estimates. Journal of Sedimentary Research, 66: 654-665.
Rice, S.P. and Church, M. (1998) Grain size along two gravel-bed rivers: statistical variation, spatial pattern and sedimentary links. Earth Surface Processes and Landforms, 23: 345-363.
Rice, S.P., Church, M., Wooldridge, C.L. and Hickin, E.J. (2009) Morphology and evolution of bars in a wandering gravel-bed river; lower Fraser river, British Columbia, Canada Sedimentology,56: 709-736.
Russell, R.D. (1939) Effects of transportation on sedimentary particles. In: Recent Marine Sediments (Eds P.D. Trask), pp. 33-47. American Association of Petroleum Geologists, Tulsa.
Sambrook Smith, G.H. and Ferguson, R.I. (1995) The gravel sand transition along river channels. Journal of Sedimentary Research Section A-Sedimentary Petrology and Processes, 65: 423-430.
Seal, R. and Paola, C. (1995) Observations of downstream fining on the North Fork Toutle River near Mount St. Helens, Washington. Water Resources Research, 31: 1409-1419.
Singer, M.B. (2008) Downstream patterns of bed-material grain size in a large, lowland alluvial river subject to low sediment supply. Water Resources Research, doi:10.1029/2008WR007183
Smith, N.D. (1974) Sedimentology and bar formation in the Upper Kicking Horse River, a braided outwash stream. Journal of Geology, 82: 205-223.
Sternberg, H. (1875) Untersuchungen über längen- und Querprofil geschiebeführender Flüsse. Zeitschrift für Bauwesen, 25: 483-506.
Surian, N. (2002) Downstream variation in grain size along an Alpine river: analysis of controls and processes. Geomorphology, 43: 137-149.
24
Whiting, P.J. (1996) Sediment sorting over bed topography. In: Advances in Fluvial Dynamics and Stratigraphy (Eds P.A. Carling and M.R. Dawson), John Wiley & Sons Ltd, Chichester.
Whiting, P.J., Dietrich, W.E., Leopold, L.B., Drake, T.G. and Shreve, R.L. (1988) Bedload sheets in heterogeneous sediment. Geology, 16: 105-108.
Wolcott, J.F. and Church, M. (1991) Strategies for sampling spatially heterogeneous phenomena: the example of river gravels. Journal of Sedimentary Petrology, 61: 534-543.
Yatsu, E. (1955) On the longitudinal profile of the graded river. American Geophysical Union Transactions, 36: 655-663.
25
Table 1. (a) Surface and (b) subsurface sediment characteristics of morphological units on
Queens bar.
Primary unit bars
Secondary unit bars
Bar-top channels
Erosional bar edge
a) Surface layer
Number of samples1 21 [19, 9] 30 [29, 14] 29 [24, 12] 7 [7, 6]
Sand (%) 6 (0-75) 1 (0-16) 17 (0-79) 9 (0-48)
Coarse-fraction2 D50 (mm) 25 (18-30) 26 (18-32) 34 (27-34) 35 (22-43)
Coarse-fraction D95 (mm) 59 (43-72) 60 (41-76) 79 (61-99) 81 (54-97)
Sorting of coarse fraction3 0.72 (0.58-0.86)
0.76 (0.62-1.00)
0.81 (0.64-1.10)
0.89 (0.56-1.02)
Structure: overloose, normal, underloose
4, 12, 5 4, 17, 9 0, 11, 18 0, 4, 3
Embeddedness: clean, slight, moderate, heavy
8, 5, 6, 2 20, 7, 3, 0 5, 10, 0, 14 0, 3, 2, 2
Imbrication: none, weak, strong, very strong
1, 9, 9, 2 1, 10, 14, 5 0, 11, 12, 6 0, 3, 3, 1
Censored surface layers?
none 5 sites none none
b) Subsurface bulk
Number of samples 1 1 1 1
Sand (%) 18 13 21 27
D50 (mm) 19 20 23 13
D95 (mm) 63 62 78 71
Armour ratio4 1.3 1.3 0.2 2.9
Entries for surface D50, D95, sorting and percentage sand are group means with minimum and maximum values in parentheses. (1) Sample size refers to the total number of samples in each group. Sorting values were calculated only for grain size distributions based on Wolman sampling and eight photographic samples that contained more than 5% sand were not used to estimate gravel D50 and D95. Values in square parentheses therefore indicate the group sizes for gravel percentile and sorting calculations, respectively. (2) Coarse-fraction refers to grain-size distributions truncated at 4mm. (3) sorting was calculated as (Ψ84 – Ψ16)/2 (Inman, 1952). Armour ratios are for the four individual sites where subsurface data are available and were calculated using corresponding untruncated surface and subsurface GSDs.
26
Figure captions:
Figure 1. Fraser River gravel reach between Laidlaw and Mission, British Columbia,
Canada. River kilometres are measured from Sand Heads (mouth of the river); the
major bars are named.
Figure 2. Supraplatform features of a large lateral compound bar (Wellington Bar) on
Fraser River.
Figure 3. Ratio of median grain sizes (no truncation) in the upstream third to downstream
third of individual lower Fraser River bars plotted against distance downstream
(measured from river km 155). Filled circles represent surface bed materials and
open circles represent subsurface materials. Points above the dashed line are those
where coarser bed materials are found toward the bar head.
Figure 4. D95 of the coarse fraction (> 4 mm) of the surface grain size distribution plotted
against site elevation for 87 samples from Queens bar. The dashed line, which has
a slope of –0.008 is an appropriate envelope that constrains the relation between
particle size and elevation for the local channel slope (see text for details).
Figure 5. Distributions of (A) coarse-fraction (> 4mm) D95 and (B) site elevation grouped
by site morphology. Queens bar sites, details in Table 1.
Figure 6. (A) Truncated (> 4 mm) subsurface grain-size distributions compared with the
surface grain-size distributions (> 4mm truncation) at the same four sites on
Queens bar. (B) Full subsurface grain-size distributions of samples collected from
each morphological setting on Queens bar.
Figure 7. Grain-size variations across unit bars and sheets: (A) discontinuous sand
deposition in the lee of a sheet crest on Spring Bar; (B) and (C) gravel bed
materials upstream and downstream of a sheet crest on Spring bar – quadrat is
0.50 by 0.50 m; (D) tail to crest fining of bed material across a series of climbing
sheets on Spring bar – circles represent coarse-fraction (> 4mm) D50 and the
locations of photographs B and C are indicated; (E) coarse-fraction (> 4mm) D95
variations and high-flow current directions across two unit bars on Queens –
dashed lines indicate avalanche faces and symbol ‘handles’ point downstream.
27
Figure 8. Flank unit bar accretion on Calamity Bar during the 2002 freshet. Total station
surveys of (A) March 2002 and (B) April 2003. Arrows indicate flow direction in
the principle channel. Colours indicate bed elevation from low in deep blue
through brown, yellow and white to green at the highest elevations.
Figure 9 Median grain size changes at reoccupied sampling positions on Queens bar
between (A) April and September 2000 and (B) September 2000 and September
2004. Sites where there was net aggradation or erosion or that are likely to have
experienced the passage of a significant sediment unit (closed circles) are
differentiated from sites where there was no morphological change (open circles).
The diagonal lines define a zone within which differences may be due to sampling
error (see text for details).
Figure 10. (A) Median surface grain size (no truncation) plotted against distance downstream
(measured from river km 155) for the entire gravel-bed reach of lower Fraser
River. Data of individual bars are indicated by the different symbols. The majority
of scatter within the general downstream fining trend is caused by within-bar
variability. Within-bar variability (between homogeneous patches) declines
downstream. (B) Median surface grain size (no truncation) plotted against
distance downstream (measured from river km 155) for the sub-reach common to
sampling programmes conducted in 2000 and 1983/84 (between river km 108 and
140). There has been some general fining, but slopes in the linear regression
models are not significantly different (Ancova, p =0.83).
Figure 11. Departures from the downstream trend correlated with bar thickness. See text for
explanations concerning the two named, anomalous bars.
28
Figure 1
29
Figure 2
30
Figure 3
31
Figure 4
32
Figure 5
33
34
Figure 7
35
Figure 8
36
Figure 9
37
Figure 10
38
Figure 11